hushem_5x_deit_tiny_adamax_00001_fold4
This model is a fine-tuned version of facebook/deit-tiny-patch16-224 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.4803
- Accuracy: 0.7857
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 1.3199 | 1.0 | 28 | 1.2249 | 0.5 |
| 1.0134 | 2.0 | 56 | 1.0887 | 0.5476 |
| 0.8677 | 3.0 | 84 | 1.0102 | 0.6190 |
| 0.708 | 4.0 | 112 | 0.9014 | 0.5952 |
| 0.5921 | 5.0 | 140 | 0.8309 | 0.6429 |
| 0.5207 | 6.0 | 168 | 0.7657 | 0.6905 |
| 0.3875 | 7.0 | 196 | 0.7178 | 0.6667 |
| 0.3518 | 8.0 | 224 | 0.6618 | 0.6667 |
| 0.2677 | 9.0 | 252 | 0.6279 | 0.7143 |
| 0.2022 | 10.0 | 280 | 0.5907 | 0.7381 |
| 0.2088 | 11.0 | 308 | 0.5564 | 0.7857 |
| 0.1641 | 12.0 | 336 | 0.5320 | 0.7857 |
| 0.1049 | 13.0 | 364 | 0.5289 | 0.7857 |
| 0.092 | 14.0 | 392 | 0.5023 | 0.8095 |
| 0.0557 | 15.0 | 420 | 0.4953 | 0.7381 |
| 0.0471 | 16.0 | 448 | 0.4998 | 0.7857 |
| 0.0348 | 17.0 | 476 | 0.4480 | 0.8095 |
| 0.0266 | 18.0 | 504 | 0.4459 | 0.7857 |
| 0.0161 | 19.0 | 532 | 0.4594 | 0.7857 |
| 0.0135 | 20.0 | 560 | 0.4976 | 0.7619 |
| 0.0093 | 21.0 | 588 | 0.4434 | 0.7619 |
| 0.0077 | 22.0 | 616 | 0.4474 | 0.7619 |
| 0.0056 | 23.0 | 644 | 0.4598 | 0.7143 |
| 0.0046 | 24.0 | 672 | 0.4362 | 0.7381 |
| 0.0037 | 25.0 | 700 | 0.4189 | 0.7857 |
| 0.0032 | 26.0 | 728 | 0.4491 | 0.7857 |
| 0.0028 | 27.0 | 756 | 0.4480 | 0.7857 |
| 0.0026 | 28.0 | 784 | 0.4540 | 0.7857 |
| 0.0022 | 29.0 | 812 | 0.4510 | 0.7857 |
| 0.0021 | 30.0 | 840 | 0.4557 | 0.8095 |
| 0.0018 | 31.0 | 868 | 0.4556 | 0.7857 |
| 0.0017 | 32.0 | 896 | 0.4590 | 0.8095 |
| 0.0016 | 33.0 | 924 | 0.4610 | 0.8095 |
| 0.0015 | 34.0 | 952 | 0.4618 | 0.8095 |
| 0.0015 | 35.0 | 980 | 0.4661 | 0.8095 |
| 0.0013 | 36.0 | 1008 | 0.4626 | 0.8095 |
| 0.0012 | 37.0 | 1036 | 0.4685 | 0.8095 |
| 0.0013 | 38.0 | 1064 | 0.4710 | 0.8095 |
| 0.0012 | 39.0 | 1092 | 0.4730 | 0.8095 |
| 0.0011 | 40.0 | 1120 | 0.4760 | 0.7857 |
| 0.0011 | 41.0 | 1148 | 0.4762 | 0.7857 |
| 0.001 | 42.0 | 1176 | 0.4741 | 0.8095 |
| 0.0011 | 43.0 | 1204 | 0.4784 | 0.8095 |
| 0.001 | 44.0 | 1232 | 0.4806 | 0.7857 |
| 0.001 | 45.0 | 1260 | 0.4792 | 0.7857 |
| 0.001 | 46.0 | 1288 | 0.4801 | 0.7857 |
| 0.001 | 47.0 | 1316 | 0.4802 | 0.7857 |
| 0.0009 | 48.0 | 1344 | 0.4803 | 0.7857 |
| 0.001 | 49.0 | 1372 | 0.4803 | 0.7857 |
| 0.001 | 50.0 | 1400 | 0.4803 | 0.7857 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0